Page 109 - Building Big Data Applications

Chapter 5 Pharmacy industry applications and usage 105

integrations and business rules applied to the data, and often we lose sight of the
number of transformations and their hidden formulas and associated rules to integrate
the data. This complexity needs to be very clearly defined and documented, especially in
the new world of data where we will be integrating the machine learning, artificial in-
telligence, and several neural network algorithms, analytics, and models. While we can
learn and fix issues as we come across them, the data volume is different, the formats are
vast and different, the infrastructure is resilient and can scale out as much needed, yet
the complexity needs to be defined for both the understanding of what is being done and
the associated outcomes.
The transformations will occur at different layers of the data architecture; we will run
transformations for operational analytics, data discovery and data exploration at the raw
data swamp layer. These transformations will be executed by multiple teams and mul-
tiple end users, and outcomes from these exercises will be validated for use of the data.
In this case the complexity is validated and will be documented if the data use case is
accepted for further analysis.
Transformations will occur further as we start the journey to data lake. This is an
enterprise asset and will be used by all users for executing the business reports,
extracting insights, and delivering more integration touchpoints. The transformations
here will be implemented as microservice architecture, which means we need to define
and design complexity within the libraries used in the microservices layers. The same
transformation exercises and integration exercises will occur in the data hub and it will
require the complexity to be broken down to manageable pieces of architecture. The
final layer of transformations is the analytical modules where we will use artificial in-
telligence, machine learning, neural network algorithms, and analytical models. These
layers deliver fantastic results but need the appropriate inputs to be applied with the
right granularity and data quality. This is another layer to manage complexity once the
data is available for compute.
Now that all possible layers and associated discussions have been had on complexity,
let us see the real-life management of data in the pharmaceutical industry.
There are several distinct use cases in the pharmaceutical industry that we will
discuss, these include the following:

Drug discovery
Patient clinical trials
Social media community
Compliance

Drug discovery is a very intricate and complex process, which needs the researchers
to develop a comprehensive understanding of how the human body works at the mo-
lecular level. This means to develop a thorough understanding of how the body reacts to
current treatments, document the in-process experiments on the different studies of
changes being developed to drugs, and have a much better grasp of the killer-effects
from consumptions of drugs including side effects and intricacies caused. All of this